claude code

Thanks But No Thanks on the Claudeswarms, Kevin Roose

by

claude code

The generosity of Kevin Roose, a New York Times tech columnist and co-host of the Hard Fork podcast, is to take pity on those who are working hard without the benefit of cloudworms.

In a January 25 post, Ruiz said he had “never seen such a deep gulf” between Silicon Valley insiders and outsiders like himself. He says the people they live with are “putting multi-agent cloudswarms in charge of their lives, consulting chatbots before every decision,” and “are wired to a degree that only sci-fi writers dare to imagine.”

The Hard Fork includes a lot of banter on Ruiz’s part – mostly directed at his more comically agile co-host Casey Newton – so it’s not wrong to me that Ruiz is trying to layer some irony and hyperbole on top of his condescension in this post. However, he quickly removes that facade in his next episode, in which he says that he “wants to believe that everyone can learn this stuff”, but he worries that perhaps, “restrictive IT policies have created a generation of knowledge workers who will never fully learn.”

Recent hard fork episodes have seen an unusual amount of excitement about vibecoding – using AI tools to do faster software engineering. Once upon a time, Github Copilot and ChatGPT wowed software engineers because they could write code like a person, and you could run the code, and the code would work. AI’s ability to code has been steadily improving since around 2021, leading some software engineers to make predictions of various forms of Armageddon.

For example, Dario Amodei, CEO of cloud parent company Anthropic, published one of these today in the form of a 38-page blog post. Amodei wrote, “Humanity is about to be entrusted with almost unimaginable power, and it is deeply unclear whether our social, political, and technological systems have the maturity to wield it.”

Roose and Newton, first and foremost, are not software engineers, but Roose recently used cloud code to build an app called Stash, an experience he talked about at Hard Fork. Stash is a read-it-later app like the defunct Pockets or the still-existing Instapaper. According to Roose, Stash does the same thing “I used to use Pocket for. Except now I own it and I can make changes in the app. And I made it, I’d say, in about two hours.” Very good. sincerely.

In another episode of Hard Fork, listeners told their stories about what they’re Vibecoded. These people probably didn’t use to code, and now they are coding, which is definitely good. One created a tool for wallpaper customers to calculate how much wallpaper they need to buy. Another created a gamification system for his kids’ homework.

With all due respect to these guys and the neat things they’re pulling off with Vibecoding, this is just people putting themselves into busywork for fun. There’s nothing wrong with that, but it is what it is.

It’s true that most people don’t have the knowledge to do software engineering work, and if like me, you’ve never coded anything, Vibecoding is interesting to try. I’ve made some rudimentary side-scrolling games out of LLM, created ray-traced 3-D environments in JavaScript, and done some other small experiments that failed. I learned a little about the LLM, but it didn’t change my life.

Then again, I, like many people, am bored with optimization and productivity hacks, and it’s not in my nature to have software ideas that are purely software. In the rare cases where I feel a creative spark that involves coding, the coding is a small part of the idea, and the rest of the idea involves a lot more engagement with the world than an LLM. For example, I live in one of those neighborhoods where people go crazy with Halloween decorations, and I’ve daydreamed about installing festive lawn animatronics, but vibecoding a control system will only get me so far in the process of configuring my monsters. Most real work I’ll be doing in my yard with a power drill, wires and poles along with my werewolf dummy, and Cloud Code is nowhere near standing that thing upright on my lawn.

Roose and other AI fanatics have been talking recently such as Its. At the end. Here. They make it seem as if AI is really going to move forward, and norms need to be enforced.

When Ruiz talks about these disenfranchised “knowledge workers” outside San Francisco, if he means specifically software engineers who are struggling to accomplish tasks that can be done by cloudswarms (in case you’re wondering, cloudswarms appear to be tiny virtual coder hives that perform complex coding tasks), then I suspect his pity is misplaced. If AI-inclined coders aren’t allowed to use the latest AI tools while on the clock, and they’re also software engineers in their spare time, it’s because they’re playing with AI toys at home if they want.

And there can be little doubt that, half-joking or not, Ruiz’s experience of people in the Bay Area “wireheading” and constantly asking chatbots for life advice is real. This is what is expected. He has a number of other problems too, like a terrible new habit of injecting himself with peptide solutions purchased online.

It’s not at all surprising that people in San Francisco think AI is going to be the closest thing possible to God, because it seems to be the closest thing to what many people in San Francisco consider God to be: a software engineer. An understandable mistake.

But the rest of the miserable knowledge workers who don’t have the privilege of living in the AI ​​paradise of San Francisco don’t necessarily believe that software engineers are that powerful, and some of us are counting down the months until next Halloween, and AI won’t do much to help our latex clowns look scary until then. This will probably never happen, and that’s okay.





<a href

Leave a Comment